← All troubleshooting topics

Sending & receiving messages

Generation runs through a single FIFO queue and Loach exposes whatever the provider reports. Most message-level oddities trace back to one of those two facts.

My new message is stuck on “waiting”

You sent a prompt in one chat but it shows a spinner without streaming anything.

Loach runs one generation at a time across all chats. If another chat is busy, your message is parked in a FIFO queue.

• Wait for the current generation to finish, or
• Open the chat you want answered and click Respond now in its header. This cancels the current runner and starts yours.

The reply stopped halfway and shows an error

A streaming reply was cancelled or errored, and the bubble ends with a red error line.

Whatever streamed before the failure is kept — open the message menu and use Copy raw if you want to preserve it. To get a full answer, send the same prompt again, or click Regenerate if it’s visible. If errors repeat, see Providers & connections or Performance & VRAM.

I see no “thinking” trace even though the model supports reasoning

A reasoning model is selected but no thinking block appears above the answer.

1. Open the Parameters sidebar and confirm the Thinking toggle is on for this chat.
2. Make sure the model actually advertises the thinking capability — the tile in the Models tab shows a “thinking” badge if it does. Many tags of the same base model differ on this.
3. Thinking is Ollama-only. OpenAI providers ignore the toggle even when it’s on.

Tokens-per-second or token count chip is missing

Some replies show a metrics chip; others don’t.

Metrics are shown when the provider reports them. Most OpenAI-compatible proxies omit the timing fields, so chats against those backends will only show the token counts (or nothing).

Replies are off-topic, repetitive, or too short

The model is technically responding but the quality is poor.

Open the Parameters sidebar:

• Lower temperature for more focused answers, raise it for more variety.
• Raise max tokens if the answer keeps cutting off.
• Raise num_ctx if the model is forgetting earlier turns. Higher num_ctx uses more VRAM.
• Increase repeat_penalty (try 1.1–1.3) if the model loops.
• Click Reset to defaults in the panel header to discard per-chat overrides and fall back to the model’s defaults.